Research map / Learning theory and algorithms / Matrix and tensor methods
WHInter: A Working set algorithm for High-dimensional sparse second order Interaction models
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use WHInter's branch-pruning bound and inverted-index working-set updates instead of SPP | LASSO with two-way interactions · Synthetic LASSO regularization paths | runtime: two orders of magnitude or more | Helped |
| Use the αℓ₂ branch bound instead of the α = 1 bound (ablation) | WHInter · Synthetic LASSO regularization paths | runtime: × 1.2 to × 1.8 | Helped |
| Use the αℓ₂ branch bound instead of the α = 1 bound (ablation) | WHInter · Chromosome 22 SNP LASSO path | runtime: × 1.4 | Helped |
| Use the αℓ₂ branch bound instead of the α = 1 bound (ablation) | WHInter · Chromosome 1 SNP LASSO path | runtime: × 1.8 | Helped |
| Use the minimizing α bound instead of the αℓ₂ bound (ablation) | WHInter · Synthetic LASSO regularization paths | runtime | Hurt |
| Use reference dual variables and maintain mref rather than the ζ bound alone (ablation) | Working-set algorithm for interaction LASSO · Synthetic LASSO regularization paths | runtime | Helped |
| Use the inverted-list update algorithm instead of SPP's safe-set identification (ablation) | Working-set algorithm for interaction LASSO · Synthetic LASSO regularization paths | runtime: × 17 faster than SPP on average | Helped |
| Use breadth-first instead of depth-first search (ablation) | SPP for pairwise interactions · SPP pairwise-interaction search | runtime: × 1.2 for n=p=1000 to × 1.6 for n=1000, p=10000 | Helped |
| Apply partial-inner-product pruning (ablation) | MIPS1 · Synthetic MIPS benchmarks | runtime: × 1.6 times faster than Naive but × 11 times slower than IL | Mixed |
| Use interaction terms rather than main effects only (ablation) | WHInter · Chromosomes 1 and 22 test sets | predictive accuracy | Helped |
| Use interaction terms rather than main effects only (ablation) | WHInter · Whole-genome test set | predictive accuracy | Hurt |
Closest papers by meaning
- SPARTAN: A Sparse Transformer World Model Attending to What Matters (NeurIPS 2025 · similarity 0.84)
- Proxy-Based Approximation of Shapley and Banzhaf Interactions (NeurIPS 2026 · similarity 0.82)
- CausalBind: Causal Modeling and Learning for Protein-Molecule Virtual Screening (NeurIPS 2026 · similarity 0.82)
- Generalist Equivariant Transformer Towards 3D Molecular Interaction Learning (ICML 2024 · similarity 0.82)
- Contrastive losses as generalized models of global epistasis (NeurIPS 2024 · similarity 0.82)
- Training Flexible Models of Genetic Variant Effects from Functional Annotations using Accelerated Linear Algebra (ICML 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.